HUMAN SUBJECT RESEARCH Rationale for Subject Selection The total number of subjects and different categories--gender, ethnicity, race, and age--continue to depend on the prevalence within the patient population availability at the Clinical Center. This protocol is a retrospective study with existing data; therefore, no new patients have been or will be recruited, and no additional imaging procedures will be performed. Risk/Benefits Analysis There is no risk to these subjects because this protocol consists of reviewing radiologic studies that have already been obtained under another IRB protocol. The potential benefit to patients is that the improved diagnosis of infectious disease will lead to advanced and earlier detection of a pathogen, leading to more prompt and effective treatment. This study is not designed to provide any direct benefits to the subjects. Consent and Documentation There is no consent form for this protocol since this study is analyzing retrospective data. Data is extracted from the CRIS, PACS, and RIS systems at an aggregate level, not at the study level. No documentation is or will be entered into the patient's medical record. ANNUAL SUMMARY The widespread prevalence and potential lethality of infectious diseases like tuberculosis, bronchiectasis, and abnormal airway enlargement pose potentially significant public health threats, and it raises the importance of needing prompt and specific diagnosis for these diseases worldwide. Diagnostic imaging--CT, MRI, nuclear medicine, ultrasound, and/or radiographic images such as mammography in an unmasked form--can possibly play a significant role in early diagnosis and in characterizing severity that can lead to morbidity and mortality. Since imaging of infectious diseases has historically been relatively nonspecific in identifying the precise pathogen, our hypothesis for this study continues to be defining imaging analysis techniques that will provide consistent qualitative visual analysis and consistent computer-aided quantitative analysis when evaluating infectious disease diagnostic studies. These new methods we are developing will soon determine if a specific methodology can increase the specificity of infectious disease imaging by using retrospective data in order to improve the prognostic value of radiology when assessing clinical severity and when enhancing the measurement of response to therapy. Since the fall of 2011, our research team retrospectively reviewed a data base of thousands of radiologic imaging studies after patients were imaged for known infections. Many of these imaging studies were categorized, examined, and analyzed by the lab staff for the identification of specific visual and quantitative imaging features to improve the accuracy of infectious disease imaging, ultimately stepping closer to a more defined patient diagnosis. Our aims were to (1) statistically evaluate all data and compare all imaging findings associated with infectious diseases, and to (2) continue to develop software algorithms that will accurately detect, quantify, and characterize infectious diseases for all diagnostic imaging studies. Using state-of-the-art computerized techniques such as graph theory algorithms and deep convolutional neural networks, we are now able to automatically and accurately evaluate imaging patterns related to various diseases. We designed and developed a series of computer tools for pulmonary CT images including delineating pathological lungs, airways, fissures, and disease-related patterns. With the help of these tools, we helped other researchers for their needs of precisely and quantitatively evaluating the progression of diseases over time.